« New at CJR: The misguided backlash against Nate Silver | Main | New at CJR: How to cover the presidential results »

November 06, 2012


"somewhat accurate" is accurate...the 90% confidence level on the establishment survey is 100,000; on the household survey its 400,000..

a disturbing number of respondents endorsed the unemployment statistics conspiracy theory. When I explicitly asked respondents whether the September figure had been manipulated for political reasons or had been calculated correctly, 55% said they thought the figures had been manipulated.

Brendan incorrectly equates conspiracy with manipulation for political reasons. It's conceivable that minor, uncoordinated manipulation by individual Dem partisans in the Labor Dept. could have changed the results by 0.2 percentage points, thus getting the figure below 8%.

In a somewhat parallel example, Glenn Reynolds observes that Yale today announced a worsening of their health plan, which was required by ObamaCare. Reynolds comments:

I think a lot of stuff will be coming out over the next few weeks and months that was carefully kept off the radar before Election Day. http://pjmedia.com/instapundit/157280/

I don't know whether Reynolds' suspicion is correct or not. My point is that he's not propounding a conspiracy. He's suggesting that there were a number of small, independent, discretioniary actions by people who wanted Obama to win.

^ What RJS Said. Did you account for that? It should have been worded with the caveat for statistics: eg beyond sampling error. Or something.

To RJS and Jedgar: Fair enough. I did not include anything about sampling error in the accuracy question, but there's no reason other than motivated reasoning to expect Rs to be more concerned about sampling error than Ds. Also, the manipulation question explicitly distinguishes between the stats being manipulated for political reasons versus calculated correctly (which implicitly allows for sampling error).

If these statistics are representative of the population of America, surely it's to be expected. A couple of weeks before the election, aren't Republican supporters going to be more suspicious of what the Democratic government is/isn't telling us than those who support the president at the moment? I would like to see how these figures differed, say, just before the 2008 election. It's likely that Obama supporters would have been suspicious of George Bush's figures (but then again, who isn't...) and McCain supporters almost entirely supportive of the government's figures. Alone, these figures show almost nothing about how information and intelligence affects conspiracy beliefs, only how a political election may persuade or even alter opinions of the reliability of the government.

Why did those with greater political knowledge give greater belief to the possibility that unemployment stats were politically manipulated? Maybe it's because this group was more aware of other cases of apparent political manipulation by this Administration. E.g., see
Foodstamps Surge By Most In One Year To New All Time Record, In Delayed Release

The comments to this entry are closed.